Program Information
Multi-View Scatter Correction for Moving-Blocker Based CBCT
C Zhao1*, X Chen2 , L Ouyang2 , J Wang2 , M Jin1 , (1) University of Texas at Arlington, Arlington, TX, (2) UT Southwestern Medical Center, Dallas, TX
Presentations
TH-AB-601-9 (Thursday, August 3, 2017) 7:30 AM - 9:30 AM Room: 601
Purpose: To develop a direct scatter estimation and adaptive edge detection method using multiple projection views for moving-blocker based cone-beam computed tomography (CBCT), which can achieve more robust and accurate scatter correction (SC) and reconstruction.
Methods: Moving-blocker based SC has been demonstrated to effectively suppress scatter contamination in CBCT. Its success highly depends on the accurate detection of blocker edges to discriminate blocked and unblocked regions in the conventional method using a single projection view. However, accurate edge detection is challenging for large size subjects. In this work, we propose to use the multiple projection views to estimate the scatter signal without edge detection at first. The lowest signal in each axial location (along the blocker moving direction) among several adjacent views is treated as scatter. The estimated scatter profiles in turn are used to adaptively determine the blocker edges, thus the primary signal in the unblocked regions for reconstruction. An anthropomorphic pelvis phantom using an Elekta XVI CBCT system was used to test the proposed multi-view SC (MVSC) method and compared with CBCT without SC and with single-view SC (SVSC).
Results: The proposed MVSC method accurately estimated the scatter signal. The adaptive thresholding based on the scatter profiles from MVSC led to robust and accurate edge detection. The reconstructed CBCT images using MVSC maintained high quality, while those without SC or using SVSC suffered severe shading or streak artifacts, respectively. Using multi-detector CT as benchmark, the root mean square error averaged in three regions of interest is 329 HU, 321 HU, and 28 HU for CBCT without SC, with SVSC, and with MVSC, respectively.
Conclusion: We developed a multi-view SC method for moving-blocker based CBCT. It is more robust to reconstruct high quality scatter corrected images than the conventional single-view SC method, especially for large size subjects.
Contact Email: